[OMPI devel] MPI_Comm_spawn[_multiple] and orted

2006-05-31 Thread Pak Lui
Hi, When I run a spawn program over rsh/ssh, I notice that each time the child program gets spawned, it will need to establish a new rsh/ssh connection to the remote node to launch orted on that node, even the parent executable and the orted are running on that node. So I wonder if there is any

Re: [OMPI devel] MPI_Comm_spawn[_multiple] and orted

2006-05-31 Thread Ralph Castain
Hi Pak I'm afraid I don't fully understand your question, so forgive me if I don't seem to address the problem adequately. As I understand it, you are asking about the scenario where someone wants to execute multiple calls of mpirun, with the applications executing on the same set of nodes. Y

Re: [OMPI devel] MPI_Comm_spawn[_multiple] and orted

2006-05-31 Thread Pak Lui
Ralph Castain wrote: Hi Pak I'm afraid I don't fully understand your question, so forgive me if I don't seem to address the problem adequately. As I understand it, you are asking about the scenario where someone wants to execute multiple calls of mpirun, with the applications executing on the

Re: [OMPI devel] MPI_Comm_spawn[_multiple] and orted

2006-05-31 Thread Ralph Castain
Pak Lui wrote: Ralph Castain wrote: Hi Pak I'm afraid I don't fully understand your question, so forgive me if I don't seem to address the problem adequately. As I understand it, you are asking about the scenario where someone wants to execute multiple calls of mpirun, with the

Re: [OMPI devel] MPI_Comm_spawn[_multiple] and orted

2006-05-31 Thread Pak Lui
Ralph Castain wrote: First, the fact that an orted already exists on a node is not sufficient to allow us to use it again for another application. The orted must be persistent or else we do not allow a new application to re-use it. This is required because the existing orted will go away when

Re: [OMPI devel] MPI_Comm_spawn[_multiple] and orted

2006-05-31 Thread Ralph Castain
Hmmm...no, it appears that we are not talking about the same problem at all. An internal comm_spawn is not at all equivalent to an external execution of another mpirun command - even though the same functions may get called, there is a very large fundamental difference between the two scenarios

Re: [OMPI devel] MPI_Comm_spawn[_multiple] and orted

2006-05-31 Thread Pak Lui
okay, sorry that I might have confused you before, but at least we are all clear about the issues. It now sounds like I'll have to try restricting the number of times the "qrsh" gets called in the sge pls module for the current release as I initially though, and revisit this issue later in ORTE